ITA-DSM: Mobile Stream Processing

The ITA-DSM project investigates the design of data stream management systems (DSMSs) for real-time data intensive applications in non-traditional environments such as mobile ad-hoc and wireless sensor networks. In these environments, connectivity to backend cloud infrastructure can be intermittent, bandwidth-constrained or in the worst case unavailable. Instead of offloading DSM to a cloud backend, the project seeks to exploit the growing computational capabilities of modern mobile phones and IoT devices by performing DSM in-situ using the combined resources of multiple devices.

In contrast to the datacentre, the major challenge for DSM in such settings is to operate over wireless networks that are highly dynamic (e.g. due to mobility or transient environmental interference). The project will therefore develop techniques to improve the adaptivity, robustness, and overall performance of DSM in wireless networks, while still providing when necessary the strong reliability guarantees of modern DSMSs for the datacentre (e.g. Twitter Storm, Apache Spark).

Funder
US Army Research Laboratory, UK Ministry of Defence, ITA (2013-2015)
Categories
Team
Dan O'Keeffe (Royal Holloway University, UK)
Theodoros Salonidis (IBM Research US)

Related Publications

Frontier: Resilient Edge Processing for the Internet of Things, O'Keeffe, Dan, Salonidis Theodoros, and Pietzuch Peter , 44th International Conference on Very Large Data Bases (VLDB), 08/2018, Rio de Janeiro, Brazil, (2018)  (1.64 MB)
Network-Aware Stream Query Processing in Mobile Ad-Hoc Networks, O'Keeffe, Dan, Salonidis Theodoros, and Pietzuch Peter , IEEE Military Communications Conference (MILCOM), 10/2015, Tampa, FL, USA, (2015)  (507.17 KB)
Outsourcing Multi-Version Key-Value Stores with Verifiable Data Freshness (Demo), Tang, Yuzhe, Wang Ting, Hu Xin, Sailer Reiner, Liu Ling, and Pietzuch Peter , IEEE International Conference on Data Engineering (ICDE), 03/2014, Chicago, IL, USA, (2014)  (188.45 KB)